Greater than 10% of all deaths are caused by cancer, therefore, it is imperative that scientific research improves and innovates the state-of-the-art in prevention, diagnosis, imaging, therapeutics, and surgery (Jemal 2011). Traditional cancer imaging techniques rely on computed tomography (CT) and positron emission tomography (PET). Both methods suffer from poor resolution and weak signal-to-noise ratios. Radioimmunoguided detection and surgery (RIGS) is a powerful modality for accurately mapping the surfaces of cancerous tissue, but the current catalog of cancer-binding antibodies are not ideal for these applications (Sun 2007).
Several generations of monoclonal antibodies have been developed against the sialyl-Tn epitope. The first two, B72.3 (Thor 1986 and Thor 1987) and CC49 (Muraro 1988; Colcher 1988), entered clinical trials for radioimmunoguided surgery (RIGS), but a significant fraction of patients developed human anti-mouse antibodies (HAMA) (Dvigi 1995). In response, a humanized variant of CC49 was constructed (AKA) (Yoon 2006; Kashmiri 1995). None of the 21 patients experience HAMA when the procedure was performed with AKA, but the third generation antibody lost more than two-fold of its binding affinity. In 2008, Yoon et al. constructed a Fab library at CDR3 of AKA (Yoon 2006). This study yielded a Fab with improved binding that was later converted to a full-length IgG named 3E8.
The latest generation of sialyl-Tn IgGs (tumor-associated antigens) are nonimmunogenic and bind the sialyl-Tn epitope with remarkable affinity. However, in order to satisfy all the requirements for imaging, these full-length antibodies require reduction to the smaller scFv scaffold. This remaining step is nontrivial, which likely describes why these imaging agents are not common place in hospitals. The variable domains are stabilized by the constant domains which are void in the truncated scFv. Independently, the VH and VL domains are only weakly associated by noncovalent interactions (and possibly disulfide bonding), thus an amino acid linker is required to assemble the full antigen binding site. Often, these engineered proteins suffer from loss of affinity, heterogeneity in quaternary structure, and diminished stability. What is needed in the art is a dramatically stabilized 3E8 antibody fragment, such as those disclosed herein.